Heart attacks are predicted with much more accuracy when done by artificial intelligence algorithm. It is a better method compared to coronary artery disease reporting and data system or any other risk assessment method. This observation has been put forward by a recent study.
The researchers explored that machine learning code finds a combination of arterial features that better-discriminated patients who did not experience Adverse event from those who did. The team developed a machine learning algorithm and trained it to use 66 CCTA image features to differentiate between living and dead patients or those having cardiac events.
They used five risk assessment methods and collected data on risk factors from CCTA studies of 6880 patients. The sample size included patients who had undergone the procedure between February 2004 and November 2009. They gave a score to the arteries of patients based on the conventional methods on one side, and the scores given by the machine learning process, on the other hand. The machine learning algorithm had also used the same data. The median follow-up time of the patients was nearly nine years.
Results of the comparison showed that the production of heart attack, mortality, myocardium infarction, and coronary heart disease death can be significantly accurate when calculated using the machine learning algorithm. It is slightly more accurate than the conventional method of assessing risk factors. For example, the area under the curve which signifies the coronary heart disease death was 0.85 for the one evaluated using the machine learning algorithm. Whereas, the area under curve evaluated by the conventional risk assessment methods was 0.79.
Hence, it is quite evident that the machine learning a logarithm is slightly more accurate compared to the risk assessment methods which were used conventionally. Similarly, the score for coronary heart disease was 0.85 used in the conventional risk assessment method and 0.80 using the machine learning algorithm.
Moreover, it has also been said that the usage of machine learning algorithm helps in predicting whether the patient should be subscribed with statins or not. The use of machine learning score indicates an accuracy of around 93% whereas the accuracy percentage of conventional risk assessment methods is just 69%.
Their words hurt and limitations in the study which were indicated by the researchers themselves. Moreover, the permutation of all the features made it possible to fully explore the modeling possibilities. Additionally, the result of myocardial infarction was underestimated because patients reply to two follow-ups were limited.
The conclusion was made that the machine learning algorithms can better predict the probability of a patient who will undergo arterial events. It can discriminate between people who are going to suffer from events and those who are not going to suffer from such events. Hence, the machine learning algorithm is much better than the conventional methods used for evaluating the chances of events.
